1 paper
Ziang Ye, Wentao Shi, Lei Chen +7
Large language model agents increasingly operate in long-horizon environments where early mistakes can trigger cascading failures and trap agents in dead-end trajectories. Agents m…